Overview

NVIDIA GeForce RTX 30 Series GPU Repair

The NVIDIA GeForce RTX 30 series — built on Ampere architecture and released from 2020 through 2022 — spanned a wide range from the RTX 3060 at the mid-range through to the RTX 3090 Ti, one of the most power-hungry consumer graphics cards ever produced at 450 watts. The lineup included the RTX 3060, 3060 Ti, 3070, 3070 Ti, 3080 (10GB and 12GB variants), 3080 Ti, 3090, and 3090 Ti.

The upper half of the RTX 30 range — 3080, 3080 Ti, 3090, and 3090 Ti — used GDDR6X memory, which runs significantly hotter than standard GDDR6. The reference cooler designs on these cards were often insufficient to keep GDDR6X junction temperatures within comfortable limits under sustained load, particularly when the thermal pads on the memory chips and backplate degraded. This produced a distinct and well-documented category of load-dependent artifacting and instability across the RTX 30 high-end range that is now one of the most common repair presentations at Brentworth for this generation.

The RTX 3060, 3060 Ti, 3070, and 3070 Ti use standard GDDR6 memory and have different thermal characteristics — their failure patterns more closely resemble typical thermal degradation and connector wear.

Ready to start?

Use the intake form to describe the device, the fault, and the result you want. The more specific you are, the easier it is to give you a useful answer.

Start Repair
Common Issues

What we fix

GDDR6X memory overheating on RTX 3080, 3080 Ti, 3090, and 3090 Ti

GDDR6X memory on the upper RTX 30-series cards runs at junction temperatures that exceed standard GDDR6 by a significant margin. The reference and many AIB cooler designs used thermal pads to transfer heat from the memory chips to the heatsink on the front and to the backplate on the rear. As these pads aged and compressed, contact quality deteriorated — producing memory junction temperatures high enough to cause instability without the GPU die temperature indicating anything was wrong. Cards that artefact or crash only during demanding workloads, while idle temperatures look normal, are the classic presentation. A full thermal pad replacement — both heatsink-side and backplate — resolves a large proportion of these cases.

Artifacting under load and memory instability

Load-dependent artefacts on RTX 30-series cards range from pixel noise and texture corruption to hard crashes and driver resets. On GDDR6X models the thermal root cause is the most common explanation. On GDDR6 models — 3060, 3060 Ti, 3070, 3070 Ti — the same symptoms can appear from general thermal paste degradation, VRM overheating, or power delivery instability. The specific model and the pattern of the artefacts matter for diagnosis — whether the fault appears immediately under load, only after extended use, or only in specific applications changes what the likely cause is.

No display output and POST failures

RTX 30-series cards that refuse to produce display output despite the system appearing to power on normally may have a VRAM fault, a display output circuit failure, a firmware corruption issue, or a power delivery problem. Cards that worked previously and developed a sudden no-display condition warrant inspection of the connectors and board before any assumptions are made. Some 3080 and 3090 cards developed VRAM faults after extended service under elevated junction temperatures — these require bench diagnosis to confirm.

RTX 3090 Ti and extreme power delivery requirements

The RTX 3090 Ti at 450 watts had one of the highest power draws of any consumer GPU produced and used a proprietary 16-pin power connector before the 12VHPWR standard was finalised. Cards at this power level are particularly sensitive to power supply quality, cable condition, and connector integrity. A 3090 Ti that crashes under load, throttles unexpectedly, or fails to sustain boost clocks should have its power delivery assessed as part of the diagnostic process.

Fan failure and cooler maintenance on Founders Edition and AIB cards

RTX 30-series Founders Edition cards use an innovative push-pull cooler design unique to that generation. AIB cards from ASUS, MSI, Gigabyte, EVGA, and others used a wide variety of triple and dual-fan cooler designs. Both Founders Edition fans and AIB fans on cards now three to five years old are encountering normal bearing wear — presenting as noise, vibration, or stall under load. Fan replacement and full cooler cleaning are recommended for any RTX 30-series card showing abnormal fan behaviour.

Common Questions

Frequently asked questions

My RTX 3080 or 3090 artefacts under load but looks fine at idle. Is it dying?

Not necessarily — this is one of the most common RTX 30-series presentations and it is frequently caused by degraded thermal pads on the GDDR6X memory chips rather than failing memory. The pads on both the heatsink-facing side and the backplate-facing side of the memory array need to maintain good contact to keep junction temperatures in range. When they degrade, memory runs hot enough to cause instability under load without the GPU core temperature showing anything alarming. A full thermal pad replacement, including backplate pads, resolves this in a significant proportion of cases. Brentworth will assess this before quoting more invasive work.

My RTX 3070 or 3060 Ti is also artifacting. Does it have the same GDDR6X problem?

No. The RTX 3070 and 3060 Ti use standard GDDR6 memory rather than GDDR6X, so they do not have the same memory junction temperature issue that affects the 3080 and 3090 range. Artifacting on GDDR6 Ampere cards is more typically caused by general thermal paste degradation on the GPU die, VRM overheating, or — less commonly — a genuine memory fault. The diagnostic and repair path for a 3070 or 3060 Ti is different from a 3090 or 3080 Ti, even when the symptoms look similar.

What is the difference between RTX 3080 10GB and 12GB for repair purposes?

The 12GB variant uses a slightly wider memory bus with additional GDDR6X chips compared to the 10GB model, but the thermal pad replacement process and diagnostic approach are the same for both. The 12GB variant was released later and has slightly less time in service, so pad degradation may be less advanced in some units. Both models are accepted for mail-in repair.

Can an RTX 3090 or 3090 Ti be repaired cost-effectively?

The RTX 3090 and 3090 Ti still carry meaningful value as high-VRAM workstation and rendering cards with 24GB of GDDR6X, which means thermal service, fan work, and connector repair are generally economically justified. Board-level microsoldering depends on the specific fault and the extent of any damage. Brentworth will provide a clear assessment of viability before any chargeable work begins.

Some RTX 3090 Ti cards used liquid metal as a thermal interface. Does that change the repair process?

Yes. Liquid metal applied at the factory on some RTX 3090 Ti AIB models requires different handling during disassembly and reassembly — it must be fully removed and correctly reapplied to avoid shorting adjacent components or creating uneven contact. If you know or suspect your card has a liquid metal thermal interface, note this in the intake form. Brentworth handles liquid metal removal and reapplication as part of thermal service on cards where it is present.

Does replacing thermal pads on an RTX 3080 void any warranty?

RTX 30-series cards are now outside their standard three-year manufacturer warranty periods in most cases. If you have an active extended warranty or retailer warranty that still applies, check the terms before sending the card for third-party repair. Brentworth does not interact with manufacturer warranty or insurance claims.

How long does RTX 30 series thermal pad replacement typically take?

Thermal pad replacement on an RTX 3080, 3090, or 3090 Ti is a multi-step process — full disassembly, cleaning the old pad and paste residue from all contact surfaces, applying new pads of the correct thickness and material to each chip position, reassembly, and bench validation under load to confirm the repair. Turn-around time varies with workload, but most straightforward thermal service cases complete within a few business days of the card arriving. Cases where the thermal assessment uncovers additional faults take longer.

Is there a diagnostic fee for RTX 30 series GPU repair?

Yes. Brentworth charges a non-refundable diagnostic fee that is credited toward the repair cost if you proceed. For RTX 30-series GDDR6X cards, the diagnostic process specifically includes thermal assessment of the memory and backplate pad condition as an early step, since this resolves a large proportion of cases before any deeper work is needed.

Ready to start?

Use the intake form to describe the device, the fault, and the result you want. The more specific you are, the easier it is to give you a useful answer.

Start Repair

Related services

Specialist Work

GPU's

Diagnosis-first GPU repair for no display, artifacting, overheating, fan failure, and damaged connectors.

View all services